Improving our understanding of the initial conditions and earliest stages of protostellar collapse is crucial to gain insight into the origin of stellar masses, multiple systems, and protoplanetary disks. Observationally, there are two complementary approaches to this problem: (1) studying the structure and kinematics of prestellar cores observed prior to protostar formation, and (2) studying the structure of young (e.g. Class 0) accreting protostars observed soon after point mass formation. We discuss recent advances made in this area thanks to (sub)millimeter mapping observations with large single-dish telescopes and interferometers. In particular, we argue that the beginning of protostellar collapse is much more violent in cluster-forming clouds than in regions of distributed star formation. Major breakthroughs are expected in this field from future large submillimeter instruments such as Herschel and ALMA.